Passing same SEED isn't generating same value as previous simulation

UVM
, , ,
1

Hi

We have a situation where we need to have same randomized values. We have same address randomized but the data field defined as random has different values in different simulations. The addrs and data are defined in two different transaction classes. Is there any possibility of having different seeds internally but we don’t see multiple seeds in the log.

2

In reply to Sv-hustler:

A UVM simulation has to generate the same randomized values in different runs. If this would not happen you could never debug your design/testbench.
Could you please show some code?

3

In reply to chr_sue:

Thanks chr_sue. I will debug the testbench to look further.

4

Some of the things that could change the random results:

  • Changing the testbench topology, especially adding new classes and new instances
  • Adding or removing random variables
  • Changing the simulator version
    Random stability tries to keep the results from changing, but if you make too many changes to the testbench, the results will differ.